continuum-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Emmanuel Venisse" <emmanuel.veni...@gmail.com>
Subject Re: ContinuumStore refactoring
Date Fri, 22 Feb 2008 09:01:12 GMT
As I already explained, I'm in favor of named queries

On Fri, Feb 22, 2008 at 2:54 AM, Rahul Thakur <rahul.thakur.xdev@gmail.com>
wrote:

> Hi,
>
> I'd like to go ahead and pick up something towards the next Continuum
> iteration. I am thinking refactoring ContinuumStore interface as was
> earlier discussed on this list and as I did on the 'continuum-jpa'
> branch.
>
> To this end, I need to get a clear picture on a few items:
>
> 1)   Which JPA provider are we going ahead with then?


I'd like to start with toplink provider.


> 2)   Criteria vs Named Queries: I am not convinced (yet) that Named
> queries are the way to go. I did some digging around, they are indeed
> best practices for JPA but I think the decision merits other
> consideration(s). I still believe the Criteria Queries will help us
> define a cleaner Store interface.


I'm always in favor of named queries.
An other point about them that I haven't explain in previous threads (I
think) is that with named queries, it is possible to modify queries
externally with xml files so if with a DB we have some performance issues,
it will be possible to override queries by a modified JPQL query or a native
query.


> 3)   Using Criteria Queries would mean that we use JPA extensions
> specific to provider. I don't see this as an issue as long as we are
> not expecting underlying JPA providers to be swapped. Thoughts?


Continuum isn't a big application and I don't think we need provider
extensions for now.

Before to start the code, I'd like to see a DB model exported to an image so
we'll can include it in the site. We must define too which datas will stay
in the db and which datas will be externalized in some files.


>
> Look forward to hear.
>
> Cheers,
> Rahul
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message